A1200 "A1230 Turbo+ II (Jaws II)" expansion

[Derek]

http://amiga.resource.cx/exp/gvp1230mk2

Hi all,

I'm new to the Classic Amiga scene after some 15 years of being away from my Amiga, in darkness. I have NO experience with expansions other than the 512kb RAM expansion in my A500 and a newly modded Gotek floppy replacement (for the same A500).

I never had an Amiga 1200 but i'm looking at building one up from parts now.

Are there any pitfalls with the kit above? I'd love to have a faster Amiga if the price is right.

[fryguy]

I wouldn't recommend that accelerator (unless you buy it with large simms included). It uses special GVP simms which is not that common.

I do believe though that i read somewhere how you could modify standard simms to be usable on it.

EDIT: yep found it here http://aminet.net/docs/hard/gvpsim64.lzh

[Derek]

Thanks Fryguy... that's way beyond my current skills i suspect. Ouch I'm utterly impressed that people are still hacking away at this old hardware though. Bloody marvellous!

Do you have a recommendation for an accelerator for an A1200? I'm a noob, having never had more than an A500 1mb RAM. So i expect myself to be a modest user of the 1200s potential. I would like to use a WiFi PCMCIA card (network moreso than internet) and a 4gb CF card as everyone seems to be doing.

Sorry, i don't mean to hijack one question into multiple: I read somewhere that some accelerators/expansion cards stop the PCMCIA from working.

[ajk]

I had that or a similar GVP accelerator back in the day, worked great. But, as mentioned, get one that already includes the RAM. Otherwise you have at least the Blizzard, Apollo and the (newly designed) ACA line of accelerators to consider.

8MB memory expansions clash with the PCMCIA slot (not physically, but they use the same memory addresses). Accelerators which have their own CPU don't have that problem.

[fitzsteve]

Either one of the new ACA cards (1220/1232) or a trusty Blizzard 1230MkIV

There are quite a few discussions about this so have a look around the boards but basically the Blizzard is the one most people will recommend, the ACA cards are obviously new with guarantee so this is a benefit to take into account.

Note once you add an accelerator you should consider an uprated PSU (converted ATX is good or one of the 4.5amp A500 PSU's)

If you notice stability issues you might need to service your Motherboard with new capacitors and in the case of the ACA Cards possibly apply timing fixes.

Again, there's lots of discussion on these boards about these subjects.

Good luck with your project.

Steve.

[fryguy]

Originally Posted by Derek

Thanks Fryguy... that's way beyond my current skills i suspect. Ouch I'm utterly impressed that people are still hacking away at this old hardware though. Bloody marvellous!

Do you have a recommendation for an accelerator for an A1200? I'm a noob, having never had more than an A500 1mb RAM. So i expect myself to be a modest user of the 1200s potential. I would like to use a WiFi PCMCIA card (network moreso than internet) and a 4gb CF card as everyone seems to be doing.

Sorry, i don't mean to hijack one question into multiple: I read somewhere that some accelerators/expansion cards stop the PCMCIA from working.

I only had Blizzard cards for my A1200 and i can recommend them.
You could also buy a new ACA card but you might have to do some timing fixes to do motherboard to make it work good (solder some SMD components).

[Derek]

Thank you all, i'll look into it

[Damion]

I have one of these with 32MB in a basic 1200 build I use for occasional gaming.

It is FLAWLESS with WHDLoad, absolutely stellar in this department, just game away without worry. Since the early 2000's I always had '060's, which can usually be made to work pretty well, but never without that occasional access fault that will ruin your game the moment you get too comfortable.

As others mentioned, make sure you get at least 8MB of RAM and you're golden. A Blizzard is still probably your best bet (just as stable, non-proprietary SIMMs), but do not be afraid of the GVP if you find one for a reasonable shake.

[Hanzu]

Originally Posted by ajk

I had that or a similar GVP accelerator back in the day, worked great.

Originally Posted by Damion

I have one of these with 32MB in a basic 1200 build I use for occasional gaming.

It is FLAWLESS with WHDLoad, absolutely stellar in this department, just game away without worry.

Hi guys!

I sorry for thread hijacking, but since the original poster has probably already made his decision. I don't want to make another thread, since this one has the exact title and some users who have experience with GVP A1230 Turbo+ II (Jaws II).

A friend of mine has had this GVP A1230 II for years. It has worked well like said here too, but after trying to use CF card with BetterWB it has suddenly become really unstable giving guru's of 8000 0004. Clean WB 3.1 install has been done 3 times without any instabilities, but clicking away or with BetterWB loaded it eventually gives this guru at some point.

Here is what is the situation now:

• Motherboard revision is Rev 2B, but we also tested and got same symptoms come with another Rev 1D3. Both have Kick 3.1 ROMs.
• All the important electrolytic capacitors are changed at Rev 2B.
• Changed PSU to "bigfoot"-clone PSU. So it is not insufficient power either.
• BetterWB runs ok without the turbo card attached. No crashes.
• No battery leaks anywhere.
• Turbo has been tested with 0MB, 4MB, 16MB, 4MB+16MB SIMM combination. With 0MB installed the turbo does not seem to guru, but screen hangs after longer period of time.
• Turbo has 68EC030 40MHZ and 68882 40MHz. Tested without FPU attached also.

In this old thread back from 1994.
https://groups.google.com/forum/#!to...re/7ixhqmQJz1s

People complain that this turbo gives gurus because Commodore produced new A1200 motherboard revisions with altered specs and GVP told some users to send these turbos back for modification to make them more compatible with later revision motherboards. I wonder how old revision is known to be compatible maybe 1D1 because 1D3 gaves same symptoms as Rev 2B.


Question 1: Is does anyone know what are these modificatons they performed for some cards or planned to perform? If you own this very same turbo, could you look for pins that look like resoldered to notice if the card has possibly been modfied by GVP. Also look for sticker texts over some programmable logic chips, so we could compare if we have same or older firmware.


Question 2: Another guestion how much would this card benefit in stability for having real 68030 50MHz instead 68EC030 40MHz which is not even gold plated when socket is and makes me believe this could result bad contact between CPU and socket in the long term?


Question 3: This card seems to have some sort of MapROM function that cannot be disabled by jumper like in Blizzard turbos. Could this patch improve stability http://aminet.net/util/misc/PatchBoot.readme and is there any reason to have ROM mapped anymore when latest Kickstart CHIPs must be with faster ns speed than the old ones?

Question 4: How important it is to have old GVP software install disks installed in WB3.1 or BetterWB with this when thinking about the stability? Maybe this site would have the latest one available http://www.gvp-m.com/updatejapan.html#Installdisks



I do not expect anyone to know all the answers, but any assumptions are welcome.

[Damion]

Question 1: Is does anyone know what are these modificatons they performed for some cards or planned to perform?

Yes! I have the service bulletin from GVP, IIRC it's the removal of two SMD resistors. I'll dig it out this evening and post the info here.

Question 2: Another guestion how much would this card benefit in stability for having real 68030 50MHz instead 68EC030 40MHz which is not even gold plated when socket is and makes me believe this could result bad contact between CPU and socket in the long term?

Well, not quite the same but I upgraded mine from full '030 40MHz rated to a Rev C 50MHz part (and 50MHz FPU). While there are a few quirks with this board, stability has been EXCELLENT. Incidentally, my card appears to have come from the factory with the 2 resistors already removed, definitely no signs of previous soldering. I had read a while back in the usenet archive that the later production runs came that way.

[Damion]

OK, found them:

Bulletin #41 - Mentions problems with the SCSI kit installed (failure to boot with the SCSI kit installed (grey screen) or no SCSI devices recognized). Solution is to remove jumpers J1 and J2 on the SCSI kit (these are termination jumpers).

Bulletin #43 - Is noted to resolve problems with later revision A1200 motherboards again in conjunction with the SCSI kit, specifically 1A. It says that the modification does not affect the performance of the 1230 with motherboards 1.A.1 or 1.D.1.

The modification is to remove the 220ohm resistors at positions R15 and R16 (again mine appears to never have had these installed).

This last bulletin mentions that removal of these parts is the only the first of "a number" of changes to improve "field performance", but gives no indication what those other changes might be. When I trawled the usenet archive the only modifications I found mention of were the removal of R15 and R16. Would be interesting to find those other service bulletins if they exist...

[Hanzu]

Thank you Damion! This unit I'm fighting has never seen SCSI module, but I believe those changes will do good even without it.

Here I see R15 and R16 removed manually:
http://amiga.resource.cx/photos/phot...res=hi&lang=en
Also FC3 (tantalum capacitor?) is left out by factory.

The soldering process quality is quite bad compared to for example to Phase 5 products. Through hole component feets seem to often solder very asymmetrick. For example See CN4 (SCSI connector pin 2)
http://amiga.resource.cx/photos/phot...res=hi&lang=en

It is also possible that all or some of the THD components were hand soldered. Because it would have created bottlenecks in production to have SMD and THD compoents on both sides.

I believe the owner of this card will go for 68030 50MHz PGA to rule out one possible suspect and to give the card more performance and value.

What motherboard REV and JAWS II REV are you using and do you have GVP software install disks installed? Would really help to have high resolution pictures of your card, to compare resistor values, sticker writing in programmable chips and so on. Clearly it seems to be a card manufactured after Bullettin #43.

If you find out anything more about the subject from those usenet archives, let us know.

I have one repair tip for anyone who has trouble with this model: check FC4 solderin in the end SIMM socket. It may break off from its solder when the card is pulled/pushed away from trapdoor and A1200 case edge pushes the capacitor back. Happened to us, but repairing that did not make the card stable.

[Damion]

Motherboard is a 1.D.1, and the card is Rev 3.04. I'm using Thor's MMULib tools for kickstart remapping, otherwise exec, expansion library and autovectors end up in chipram. Still fiddling with the software, can't seem to "loadmodule" Piru's exec correctly (yet).

The soldering looks pretty good on my card, I imagine there must have been some production variations. I'll try to get some pictures taken.

Question: Would you mind booting to WB, running a "bustest chip" before removing the resistors? Any non-laced screenmode with 8 or less colors should suffice. :-)

[Damion]

I added the resistors, here the card is only stable without them. Benchmarks are ultimately identical with a few minor oddities that stabilize after 10-15 minutes of warm-up.

For fun I tried a 56.66MHz oscillator, it appears rock solid, but only without the resistors. I ran AIBB and a bunch of demos requiring an FPU, no problem. I put it back to 50 before closing her up since I'm too chicken to overclock old hardware these days. :P

The card writes to chipram a bit slower than the Blizzard 1230-IV and Apollo cards, but it's great for WHDLoad. 56MHz gives you about 10-13% boost, which roughly puts you 5-10% faster than the 50MHz Blizz in the AIBB tests. :-)


*edit*

Just saw this and wanted to respond:

Originally Posted by Hanzu

Question 3: This card seems to have some sort of MapROM function that cannot be disabled by jumper like in Blizzard turbos. Could this patch improve stability http://aminet.net/util/misc/PatchBoot.readme and is there any reason to have ROM mapped anymore when latest Kickstart CHIPs must be with faster ns speed than the old ones?

The GVP's MapROM function is only activated by the software tool (gvpcpuctrl), default is disabled. (By contrast, the Blizzard maps KS at boot unless disabled by the jumper.) Unless I'm mistaken, I believe the whole point was to have KS mapping when using mmu-less (EC) CPUs.

With regard to whether or not remapping is still beneficial:

4.Ram Disk:> bustest rom
BusSpeedTest 0.19 (mlelstv) Buffer: 262144 Bytes, Alignment: 32768
========================================================================
memtype addr op cycle calib bandwidth
rom $00F80000 readw 365.4 ns normal 5.5 * 10^6 byte/s
rom $00F80000 readl 364.9 ns normal 11.0 * 10^6 byte/s
rom $00F80000 readm 329.1 ns normal 12.2 * 10^6 byte/s

And again with remapping enabled:

4.Ram Disk:> gvpcpuctrl fastrom=on

GVPCpuCtrl V5.0 ©1993 GVP
-------------------------
Kickstart ROM is now mapped to 0x02f80000.

4.Ram Disk:> bustest rom
BusSpeedTest 0.19 (mlelstv) Buffer: 262144 Bytes, Alignment: 32768
========================================================================
memtype addr op cycle calib bandwidth
rom $00F80000 readw 185.7 ns normal 10.8 * 10^6 byte/s
rom $00F80000 readl 185.1 ns normal 21.6 * 10^6 byte/s
rom $00F80000 readm 154.4 ns normal 25.9 * 10^6 byte/s

I don't get the problems PatchBoot is supposed to fix regardless of using gvpcpuctrl or mufastrom.

[tom256]

Originally Posted by Damion

I added the resistors, here the card is only stable without them. Benchmarks are ultimately identical with a few minor oddities that stabilize after 10-15 minutes of warm-up.

For fun I tried a 56.66MHz oscillator, it appears rock solid, but only without the resistors. I ran AIBB and a bunch of demos requiring an FPU, no problem. I put it back to 50 before closing her up since I'm too chicken to overclock old hardware these days. :P

The card writes to chipram a bit slower than the Blizzard 1230-IV and Apollo cards, but it's great for WHDLoad. 56MHz gives you about 10-13% boost, which roughly puts you 5-10% faster than the 50MHz Blizz in the AIBB tests. :-)

About overclocking MC68030:
With Blizzard IV and 50ns RAM I was able to reach 62.5MHz with MC68030 mask F91C, tested on 2 Blizzards cards and 2 mask F91C.
I tried also two MC68030 mask G40W, but it won't run at 62,5MHz. Anyway mask G40W is running fine at 56MHz on ACA1230 card, so still 10% OC. Mask F91C FTW
What is funny, none of Blizzards run with 60MHz clock :P None of CPU hit 65MHz.

I still have for testing Jaws II and E-Matrix.


Tested with Heroes 2 from meynaf, because its loading CPU much. Radiator is very needed.

[Hanzu]

Originally Posted by Damion

The soldering looks pretty good on my card, I imagine there must have been some production variations. I'll try to get some pictures taken.

Question: Would you mind booting to WB, running a "bustest chip" before removing the resistors? Any non-laced screenmode with 8 or less colors should suffice. :-)

This card is REV.3 not REV.3.04 like yours. So we know now for sure there are different revisions. REV.3 still had R15 and R16 resistors installed. Here are weak results with R15 and R16:

BusSpeedTest 0.19 (mlelstv) Buffer: 262144 Bytes, Alignment: 32768
========================================================================
memtype addr op cycle calib bandwidth
chip $00078000 readw 904.5 ns normal 2.2 * 10^6 byte/s
chip $00078000 readl 903.5 ns normal 4.4 * 10^6 byte/s
chip $00078000 readm 948.5 ns normal 4.2 * 10^6 byte/s
chip $00078000 writew 908.9 ns normal 2.2 * 10^6 byte/s
chip $00078000 writel 909.6 ns normal 4.4 * 10^6 byte/s
chip $00078000 writem 909.7 ns normal 4.4 * 10^6 byte/s

Here are results with R15 and R16 removed:

BusSpeedTest 0.19 (mlelstv) Buffer: 262144 Bytes, Alignment: 32768
========================================================================
memtype addr op cycle calib bandwidth
chip $00078000 readw 904.5 ns normal 2.2 * 10^6 byte/s
chip $00078000 readl 903.5 ns normal 4.4 * 10^6 byte/s
chip $00078000 readm 948.5 ns normal 4.2 * 10^6 byte/s
chip $00078000 writew 908.9 ns normal 2.2 * 10^6 byte/s
chip $00078000 writel 909.6 ns normal 4.4 * 10^6 byte/s
chip $00078000 writem 909.7 ns normal 4.4 * 10^6 byte/s

Both results run with high resolution non-interlaced and only 8 colors.

No increase in stability or performance. This system crashes often when using Ed (to edit) or running Octamed player. Games it for example Turrican it seems to run stable. So to know more have to try something else too than BetterWB installation.

Originally Posted by tom256

About overclocking MC68030:
With Blizzard IV and 50ns RAM I was able to reach 62.5MHz with MC68030 mask F91C, tested on 2 Blizzards cards and 2 mask F91C.
I tried also two MC68030 mask G40W, but it won't run at 62,5MHz. Anyway mask G40W is running fine at 56MHz on ACA1230 card, so still 10% OC. Mask F91C FTW
What is funny, none of Blizzards run with 60MHz clock :P None of CPU hit 65MHz.

I still have for testing Jaws II and E-Matrix.


Tested with Heroes 2 from meynaf, because its loading CPU much. Radiator is very needed.

Nice to know. Waiting for 68030 50MHz PGA from Ebay to arrive. Is there a list of 68030 PGA masks somewhere and which one is the last ever produced?


The original battery and oscillator seem to heat up from the surface. I guess they absorb the heat from PCB because they have so much metal. Do you experience the same?

[tom256]

Originally Posted by Hanzu

This card is REV.3 not REV.3.04 like yours. So we know now for sure there are different revisions. REV.3 still had R15 and R16 resistors installed.

My Jaws II is rev 3 and has no R14,R15 neither FC3.

Originally Posted by Hanzu

Nice to know. Waiting for 68030 50MHz PGA from Ebay to arrive. Is there a list of 68030 PGA masks somewhere and which one is the last ever produced?


The original battery and oscillator seem to heat up from the surface. I guess they absorb the heat from PCB because they have so much metal. Do you experience the same?

MC68030 has revision A, B, and C. Processor with re. C was produced also by Freescale in 0,5um. (probably rumor only)

MC68RC030 version 16,20,25,33,40,50MHz
C74N 1um
D66C 1um
F91C rev 'C' 0,8um

F91C was prodused in 1996 and maybe 1997. Rev G40W was produced in 1997.
I'v seen also mask D66C in PLCC package produced by Freescale.


68EC030 version 25MHz, 40MHz

68EC030 G40W rev 'c' - 0,8um (earlier rev was 1um)

non offcial mask (no docs):
MC68030 mask G80W from 2004 year, 42 week
Also I'v seen mask G70W but no offical info.

[Damion]

Same bustest numbers here on the GVP. I have an Apollo 1230 MK2 that does much better (up to 7 MB/s read/write) but after testing a zillion games, I can only find a very few cases where the GVP is noticeably a hair slower. And ofc, while very fast the Apollo is not fully stable.

Anyhow, if games are working well, I bet the problem is in the OS install, HD settings or something. Another good test is Clickboom's Quake, though unplayable on the '030 it still sorts the wheat from the chaff where stability is concerned. (Here the GVP can run it forever, where the Apollo b0rks within a minute or two.)

[Hanzu]

Originally Posted by Damion

Anyhow, if games are working well, I bet the problem is in the OS install, HD settings or something. Another good test is Clickboom's Quake, though unplayable on the '030 it still sorts the wheat from the chaff where stability is concerned. (Here the GVP can run it forever, where the Apollo b0rks within a minute or two.)

We have now concluded it is not OS install. It is the GVP 1230+ II itself.

• BetterWB runs stable without GVP.
• AmigaOS 3.1 runs stable without GVP.
• Some floppy or WHDoad games run stable without GVP.

All will hang or guru when GVP is installed, even without any IDE cables hooked up and also with another Bigfoot PSU so that is not the issue either. AmigaOS 3.1 crashes much faster than BetterWB, after just opening some folder or top bar.

Not mentioned in all found documentation, but JP1 seems to disable GVP card CPU and memory and use A1200 motherboard CPU. Will test that now.

50MHz CPU has not yet arrived to test with.

Where can I buy or find clickBOOMQuake? It is not available on clickBOOM or on any Amiga resellers.

[Hanzu]

Originally Posted by tom256

MC68030 has revision A, B, and C. Processor with re. C was produced also by Freescale in 0,5um. (probably rumor only)

MC68RC030 version 16,20,25,33,40,50MHz
C74N 1um
D66C 1um
F91C rev 'C' 0,8um

F91C was prodused in 1996 and maybe 1997. Rev G40W was produced in 1997.
I'v seen also mask D66C in PLCC package produced by Freescale.


68EC030 version 25MHz, 40MHz

68EC030 G40W rev 'c' - 0,8um (earlier rev was 1um)

non offcial mask (no docs):
MC68030 mask G80W from 2004 year, 42 week
Also I'v seen mask G70W but no offical info.

68030 50MHz (mask C74N) arrived, but GVP 1230 Turbo+ II Rev.3 is still unstable with it. I think I will do timing fixes for the mobo to see if they help.